def call_samtools(file, tempdir):
'''
run samtools to generate sorted BAM files.
'''
logging.info("Running Samtools sort...")
bam_path = file[:-3] + "bam"
cmd = ("samtools view -Sb %s | samtools sort -@ 4 -O bam -T %s - > %s" % \
(file, tempdir, bam_path))
r, o, e = exe(cmd)
if r != 0:
logging.error("Samtools sort failed!")
logging.error("RETCODE %d" % (r))
logging.error("STDOUT %s" % (str(o)))
logging.error("STDERR %s" % (str(e)))
logging.error("Exiting")
exit(r)
logging.info("Finished Samtools sort.")
logging.info("Running Samtools index...")
cmd = ("samtools index %s" % (bam_path))
r, o, e = exe(cmd)
if r != 0:
logging.error("Samtools index failed!")
logging.error("RETCODE %d" % (r))
logging.error("STDOUT %s" % (str(o)))
logging.error("STDERR %s" % (str(e)))
logging.error("Exiting")
exit(r)
logging.info("Finished Samtools index.")
return bam_path
def check_bai(file, tempdir):
'''
check the index of a BAM file.
'''
if os.path.exists(file + ".bai"):
logging.info("The bam file is legal.")
return ""
else:
logging.info("The bam.bai is missed.")
logging.info("Running Samtools sort...")
bam_path = file[:-3] + "sorted.bam"
cmd = ("samtools sort -@ 4 -O bam -T %s -o %s %s" % (tempdir, \
bam_path, file))
r, o, e = exe(cmd)
if r != 0:
logging.error("Samtools sort failed!")
logging.error("RETCODE %d" % (r))
logging.error("STDOUT %s" % (str(o)))
logging.error("STDERR %s" % (str(e)))
logging.error("Exiting")
exit(r)
logging.info("Finished Samtools sort.")
logging.info("Running Samtools index...")
cmd = ("samtools index %s" % (bam_path))
r, o, e = exe(cmd)
if r != 0:
logging.error("Samtools index failed!")
logging.error("RETCODE %d" % (r))
logging.error("STDOUT %s" % (str(o)))
logging.error("STDERR %s" % (str(e)))
logging.error("Exiting")
exit(r)
logging.info("Finished Samtools index.")
return bam_path
def load_sam_multi_processes(args):
'''
task scheduling
'''
temporary_dir = args.temp_dir if args.temp_dir.endswith('/') else \
"%s/"%(args.temp_dir)
os.mkdir("%ssignatures" % temporary_dir)
# Major Steps:
# loading alignment file: bam format
samfile = pysam.AlignmentFile(args.input)
# loading reference genome
Ref = load_ref(args.Reference)
global_ref.append(Ref)
# acquire the total numbers of the ref contigs
contig_num = len(samfile.get_index_statistics())
logging.info("The total number of chromsomes: %d" % (contig_num))
# Thread scheduling
process_list = list()
for i in samfile.get_index_statistics():
process_list.append([i[0], i[3]])
# #chr #read
process_list = sorted(process_list, key=lambda x: -x[1])
# start to establish multiprocesses
analysis_pools = Pool(processes=args.threads)
# Acquire_Chr_name
for i in process_list:
para = [(temporary_dir, i[0], args.input, args.min_distance, \
args.min_support, args.min_length)]
analysis_pools.map_async(multi_run_wrapper, para)
analysis_pools.close()
analysis_pools.join()
samfile.close()
output_p = args.output_dir if args.output_dir.endswith('/') else \
"%s/"%(args.output_dir)
if not os.path.exists(output_p):
os.mkdir(output_p)
merge_cmd = ("cat %ssignatures/* > %spotential_ME.fa" %
(temporary_dir, output_p))
r, o, e = exe(merge_cmd)
if r != 0:
logging.error("Merging ME signatures failed!")
logging.error("RETCODE %d" % (r))
logging.error("STDOUT %s" % (str(o)))
logging.error("STDERR %s" % (str(e)))
logging.error("Exiting")
exit(r)
logging.info("Cleaning temporary files.")
cmd_remove_tempfile = ("rm -r %ssignatures" % (temporary_dir))
r, o, e = exe(cmd_remove_tempfile)
if r != 0:
logging.error("Cleaning temporary files failed!")
logging.error("RETCODE %d" % (r))
logging.error("STDOUT %s" % (str(o)))
logging.error("STDERR %s" % (str(e)))
logging.error("Exiting")
exit(r)
def call_ngmlr(inFile, ref, presets, nproc, outFile):
"""
run ngmlr to generate alignments
"""
outFile = outFile + "map.sam"
logging.info("Running NGMLR...")
cmd = ("ngmlr -r %s -q %s -o %s -t %d -x %s" % (ref, inFile, outFile, \
nproc, presets))
r, o, e = exe(cmd)
if r != 0:
logging.error("NGMLR mapping failed!")
logging.error("RETCODE %d" % (r))
logging.error("STDOUT %s" % (str(o)))
logging.error("STDERR %s" % (str(e)))
logging.error("Exiting")
exit(r)
logging.info("Finished NGMLR mapping.")
return outFile
def call_ngmlr(inFile, ref, presets, nproc, outFile, SUBREAD_LENGTH,
SUBREAD_CORRIDOR):
"""
fq = input file
automatically search for .sa
"""
outFile = outFile + "cluster.sam"
logging.info("Running NGMLR...")
cmd = ("ngmlr -r %s -q %s -o %s -t %d -x %s --subread-length %d --subread-corridor %d" \
% (ref, inFile, outFile, nproc, presets, SUBREAD_LENGTH, SUBREAD_CORRIDOR))
r, o, e = exe(cmd)
if r != 0:
logging.error("NGMLR mapping failed!")
logging.error("RETCODE %d" % (r))
logging.error("STDOUT %s" % (str(o)))
logging.error("STDERR %s" % (str(e)))
logging.error("Exiting")
exit(r)
logging.info("Finished NGMLR mapping.")
return outFile